Our long term goal is to understand the functional architecture of neural circuits in cat retina. Here we focus on the cone circuits to on-beta (X) and on-alpha (Y) ganglion cells. These types (with their off counterparts) account for greater that 90% of the optic nerve cross- section and divide spatio-temporal frequency space. Beta cells (narrow field, densely distributed) serve higher spatial and lower temporal frequencies; alpha cells (wide field, sparsely distributed) serve the reverse. Thus far, we have determined for both types the complete sets of parallel bipolar circuits that connect them to rods and cones. This has led to specific hypotheses regarding the basis for receptive field structure at different luminances. We now propose: 1) Complete the structure of the beta circuit. Determine amacrine cell types that supply the last 60 unidentified synapses; quantitate cone contacts to the three bipolar types with input to the beta cell (intracellular filling + confocal and electron microscopy). 2) Probe function at outer plexiform by localizing molecules that serve GABA and glutamate transmission. For GABA, localize the synthetic enzyme, transporter, and specific subunits of the GABAA receptor. For glutamate, localize (in on bipolar cells) components of G protein system possibly corresponding to the phototransduction cascade: transducins, phosphodiesterase, arrestins, cGMP ion channels (immuno- and in situ hybridization histochemistry). 3) Determine how known circuit gains for alpha and beta relate to the number of bipolar-to-ganglion cell synapses per retinal area. Measure bipolar synapses/ganglion cell membrane area at different eccentricities (EM) and dendritic membrane area/retinal area (fill + confocal). 4) Determine how the unique 3-D architectures of the alpha and beta dendritic meshworks are accounted for by known functional constraints (such as density, coverage, membrane area/retinal area, dendritic volume) and constraints on dendritic spacing. Quantitate these factors (fill neighbors of same type, confocal + 3-D graphics). 5) Assess how identified factors (optical + neural) account for beta ganglion cell performance. Construct compartmental models and ideal observer models; compare these to actual performance; assess which factors contribute to the difference. The proposed studies address fundamental mechanisms responsible for "image processing" by identified circuits in mammalian brain. These studies will help to understand human retina because many features of retinal organization are shared across species.